Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces the Evidence-Grounded Terminology Adaptation (EGTA) framework, which enhances simultaneous speech translation (SimulST) by leveraging a document terminology memory to selectively inject relevant technical terms based on the current streaming context. Through diagnostic experiments, the authors demonstrate that the primary gains in translation quality stem from improved terminology recovery rather than broad semantic enhancements. Evaluations on two technical talk datasets reveal significant improvements in BLEU scores and recall metrics for named entities and acronyms, showcasing EGTA's effectiveness across various SimulST settings without requiring full model fine-tuning.
Contextualizing technical terminology can boost translation accuracy by over 79% in named-entity recall without the need for extensive model retraining.
Extra context is valuable for simultaneous speech translation of technical talks, but injecting the entire document context into every streaming segment is often too coarse. Through diagnostic experiments, we find that context gains mainly come from paper-specific terminology recovery rather than uniform semantic enhancement. We therefore propose EGTA, an Evidence-Grounded Terminology Adaptation framework that builds a document terminology memory, selects compact candidate terms conditioned on the current streaming state, and adapts ASR/speech-side and decoder-side decision spaces using only the selected terms. EGTA can be instantiated in cascaded, end-to-end, and generation-only SimulST settings without full-model fine-tuning. We evaluate EGTA on an ACL technical-talk SimulST evaluation suite consisting of MCIF-dev and ACL60/60-dev. On MCIF-dev, EGTA-RG improves BLEU by +1.05/+0.59, XCOMET-XL by +0.019/+0.006, named-entity recall by +79\%/+73\% relative, and acronym recall by +0.099/+0.171 on En$\rightarrow$Zh and En$\rightarrow$De. Across MCIF-dev latency settings, EGTA consistently improves XCOMET-XL, named-entity recall, and acronym recall. External validation on ACL60/60-dev further shows consistent terminology-recall gains without additional fine-tuning. Shuffled-memory controls and activation audits provide evidence that the improvements are tied to paper-specific evidence alignment rather than generic context prompting.